
By Miguel Munoz – Global rail director – Expleo Group
Signalling is already highly software-driven, but despite this, engineering productivity is being constrained by a surplus of work that gets repeated from one project to the next.
The challenge is not necessarily the technology itself, however, but how we structure and deliver the engineering work. Signalling engineering has become increasingly software and data driven, while some of our engineering approaches have yet to fully evolve at the same pace.

It’s time for change
Requirements, specifications, documentation, and configurations currently create the most friction for signalling engineers.
Picture the scenario: a team starts work on a new project that shares many similarities with one already delivered. The specific requirements are different, as every project has its own operational, technical, and regulatory context – but many of the underlying engineering challenges, configurations, test activities, and documentation may already have been addressed in some form.
This is where the next productivity opportunity lies: industrialising engineering through data and AI. Every project generates a wealth of engineering knowledge, requirements, configurations, test cases, design decisions, and documentation. If we capture and structure that data effectively, AI can help us reuse what has already been learned, adapt proven solutions to new project requirements, and continuously improve how we engineer signalling.
The objective is not to make every project the same, though, but to make every new project smarter and more efficient than the last.
And the need for this approach is only becoming greater as signalling technologies and architectures continue to evolve. Technologies such as Future Railway Mobile Communication System (FRMCS), the European Train Control System (ETCS), and cloud computing, together with the growing push for standardisation, are reshaping what future signalling programmes need to deliver.
At the same time, we are starting to see the foundations of this more industrialised approach emerging. Digital twins and Model-based design are bringing us closer to a way of working where more of the system can be modelled digitally, allowing validated specifications and test scenarios to be carried forward and adapted to new projects rather than starting again each time.
Physical access to railway infrastructure is expensive, constrained, and often difficult to schedule. The more testing and assurance that can be carried out virtually, the less dependent development becomes on scarce physical access.
AI can take this a step further, helping engineers simulate, automate, reuse, and build on existing engineering work. But there is an important limit: signalling is safety-critical engineering. If AI generates a requirement, test case, or piece of documentation, it must still be rigorously assured by an engineer and validated against the applicable railway safety requirements.
The opportunity, therefore, is not to replace engineering judgement, but to augment it using data, models, and AI; to make engineers more productive while keeping safety and validation firmly at the centre.
That means you’re not going to get an interlocking fully developed by AI any time soon, and nor should that be the immediate goal. The more immediate opportunity is to use AI around the safety-critical core: reducing repetitive engineering work while leaving responsibility for verification and validation with the engineer.
The productivity problem
Every railway project has its own operational, technical, and regulatory requirements. The opportunity is to make every new project smarter by building on what has already been learned.
That means moving beyond a purely project-by-project approach towards a more industrialised model of signalling engineering – one where data, AI, model-based design, and simulation allow proven engineering knowledge to be captured, reused, and adapted to each new project.
AI will not replace the engineer, particularly in safety-critical signalling. But it can reduce repetitive engineering work, help teams make better use of knowledge from previous projects, and support faster testing and validation.
That is where the next significant productivity gain will come from: not asking engineers to work harder, but giving them better tools, better data, and better ways to simulate, reuse, test, and validate their engineering.
The goal is simple: make every project more efficient, more connected, and smarter than the last.
Ready to rethink your approach to signalling engineering?
Discover how Expleo’s rail engineering services can support your move towards software-defined, integrated systems.





